Routers and switches deployed in packet networks consist of many hardware components. Reliability of each hardware component is characterized by predicted Mean Time Between Failures (MTBF) provided by vendor. In large networks the frequency of replacements for a given component can be used to calculate the field MTBF for some time period T (e.g., one month). The service provider can compare the field MTBF (actual MTBF of the component as observed in the field) and the predicted MTBF and request corrective actions from the vendor if the field MTBF is below the predicted MTBF. Repeated component failures with automatic recovery and software failure are not counted in the field MTBF although these failures are impacting customers. Therefore, the field MTBF metric does not accurately reflect the reliability of network service supported by the underlying network equipment.